The Complete Guide to Using AI in the Government Industry in New Zealand in 2025

New Zealand's 2025 AI moment is a policy-and-practice push: the government's OECD-aligned national AI Strategy and the February Public Service AI Framework set a light‑touch, principles‑based path that prioritises safe adoption, Treaty of Waitangi inclusion and practical productivity gains rather than heavy-handed rules - because AI here is framed as an economic tool, not just tech for tech's sake.

With officials pointing to as much as NZ$76 billion of upside by 2038, ministries are pairing guidance with capacity (new supercomputing and Public Research Organisations) while businesses race to catch up: over 82% of organisations reported AI use in 2025 and 93% say it boosted efficiency.

Barriers remain - SME hesitancy, skills gaps and privacy reform - so public servants need clear, hands-on training; practical courses like Nucamp's AI Essentials for Work offer prompt‑writing and workplace AI skills to turn strategy into day‑to‑day improvements.

Read the full strategy and guidance at Nemko and the productivity trends at Kinetics.

“Adopting generative AI alone could add NZ$76 billion (US$45 billion) to the New Zealand economy by 2038, or over 15 percent of our GDP.”

New Zealand's policy stance in 2025 favours a light-touch, principles‑based path that leans on international norms and existing statutes rather than creating a brand‑new AI Act: Cabinet adopted the OECD AI Principles as guiding values in June 2024, and the Government pairs a proportionate, risk‑based regulatory stance with practical guidance so agencies and businesses can move confidently (see DLA Piper's clear breakdown of the AI Strategy).

That means using the Privacy Act 2020, Fair Trading Act 1986, Companies Act 1993 and other existing laws - alongside Treaty of Waitangi obligations - to manage harms while keeping innovation flowing, and embedding the Public Service AI Framework's human‑centred principles and six work‑programme pillars across the public sector.

The emphasis on adoption over heavyweight rule‑making encourages agencies to test proven solutions, scale responsibly, and build capability; a memorable reminder of why oversight matters came when the Strategy - prepared with AI assistance - included an apparent AI “hallucination” misnaming the Commerce Act, showing human review is non‑negotiable.

For the ethical foundation, the OECD's updated AI Principles remain the touchstone for trustworthy, interoperable governance.

“The time has come for New Zealand to get moving on AI.” - Minister Shane Reti

For teams needing a quick map of what to read and when: New Zealand's AI story moves from principle to practice - Cabinet adopted the OECD AI Principles in June 2024, the Public Service AI Framework landed in February 2025 to steer government use, and on 8 July 2025 the Government published New Zealand's first national AI Strategy alongside a practical “Responsible AI Guidance for Businesses” to accelerate safe uptake; the official strategy and guidance are available from MBIE for download and action.

These documents collectively emphasise adoption over heavy‑handed rule‑making (New Zealand was the last OECD country to publish a national AI strategy), encourage reliance on existing laws like the Privacy Act 2020, and show the public sector “walking the talk” - the strategy itself was prepared with AI assistance as a demonstration of practice plus human oversight.

Public servants should bookmark the Public Service AI Framework on the Government Chief Digital Officer site to align departmental projects with cross‑agency standards, and use the MBIE guidance as the starting point for procurement, governance and sandbox testing before scaling.

“The time has come for New Zealand to get moving on AI.” - Minister Shane Reti

Legal touchpoints for New Zealand government AI projects start with the Privacy Act 2020 and its 13 Information Privacy Principles (IPPs): collect only what's necessary, be transparent, keep data secure, minimise retention, and treat cross‑border transfers with care - all overseen by the Privacy Commissioner (see the full Privacy Act 2020 (New Zealand) - full legislation text).

Agencies must appoint a privacy officer, treat notifiable privacy breaches seriously (the guidance points to swift notification to the Commissioner and affected people - think of the 72‑hour clock after a serious breach), and use Privacy Impact Assessments and human oversight when deploying models.

Cross‑border rules and contractual safeguards are essential (the Commissioner can issue transfer‑prohibition notices), while sector codes - including an emerging biometric code - tighten rules for specific data types.

The Office of the Privacy Commissioner's AI guidance flags transparency, accuracy, accountability and te ao Māori considerations, so platform choice, data residency and enterprise‑grade contracts matter in practice; follow the Commissioner's AI guidance and document senior‑level approvals before scaling.

The Public Service AI Framework - a non‑binding, principles‑based playbook published in early 2025 and led by the Government Chief Digital Officer - is designed to steer agencies toward responsible, productivity‑focused use of AI so government can modernise services without losing public trust; it deliberately aligns with the OECD AI Principles and sets five core AI values alongside six practical work‑programme pillars to make adoption safer and more consistent across departments.

The Framework pairs human‑centred rules (transparency, accountability, privacy and security) with an implementation agenda - governance, guardrails, capability, innovation, social licence and New Zealand's global voice - so agencies can experiment within clear boundaries and reap benefits like reduced wait times and faster triage of citizen queries.

For a concise legal and practical walkthrough see DLA Piper's explainer on the Framework and the Government's Guidance for safe use of AI in the public sector for operational expectations.

“Use of AI technologies to improve public services is a priority for me, and this guidance will enable its safe and responsible uptake.” - Hon Judith Collins KC

AI has moved from experiment to everyday tool across Aotearoa but adoption and readiness sit unevenly: surveys show 82% of New Zealand organisations were using AI in 2025 and 93% reported efficiency gains, while large firms climbed from roughly 48% in 2023 to about 67% in 2024 - yet small and medium enterprises lag, with around 68% saying they had no plans to adopt AI, a gap that risks leaving practical gains untapped (see the Kinetics AI-driven productivity gains report for New Zealand 2025 and DLA Piper's analysis of New Zealand's AI strategy).

The blockers are familiar and fixable: stubborn skills shortages (many non‑users cite lack of expertise), patchy data quality and legacy systems, budget limits, and surprisingly weak governance - only a small fraction of users have audit assurance, formal policies or staff training in place.

That combination means many organisations are getting quick wins from automation and analytics, but without stronger guardrails and upskilling the benefits may be uneven across regions and sectors; the Government's Strategy and industry guides aim to turn that momentum into lasting capability (see the Datacom State of AI Index for practical detail on governance gaps and rapid recent uptake).

“It's positive to see New Zealand businesses realising the benefits of AI and feeling more confident in their understanding of AI risks and opportunities.” - Justin Gray, Datacom

Practical first steps for government teams are intentionally modest and measurable: start by defining a clear purpose for any AI pilot and align it with the Public Service AI Framework and the Government's Responsible AI Guidance so projects solve a real pain (for example, a triage bot that trims queuing and frees frontline staff to focus on complex cases).

Map data flows against the Privacy Act and run an Algorithm Impact Assessment early - the Algorithm Impact Assessment user guide is a practical, question‑by‑question checklist to spot risks before code ships - then scope a small sandbox pilot with senior sign‑off and human‑in‑the‑loop controls.

Set governance from day one (a nominated privacy officer, a compliance owner and procurement checks), require documented testing, bias‑checks and monitoring, and build a focused upskilling plan so staff can own oversight.

Treat procurement as regulatory design: choose vendors that support data minimisation and cross‑border safeguards, iterate quickly on evidence from pilots, and scale only when monitoring shows safe, equitable outcomes; Bell Gully's explainer on NZ's light‑touch approach highlights the need to pair adoption with governance.

These steps turn the country's principles‑based stance into practical, low‑risk wins for New Zealanders. New Zealand Government guidance for safe use of AI in the public sector · New Zealand Algorithm Impact Assessment user guide · Bell Gully analysis of New Zealand's AI strategy and opportunities for business

“Use of AI technologies to improve public services is a priority for me, and this guidance will enable its safe and responsible uptake.” - Hon Judith Collins KC

Sector priorities in New Zealand in 2025 target the places where AI delivers the clearest productivity wins - agriculture, healthcare, finance, manufacturing and core public services - while recognising certain uses are higher risk and need extra guardrails.

In agriculture the focus is precision farming and agritech (think AI collars for herd management and autonomous orchard vehicles) to raise yields and cut inputs, but data quality, rural connectivity and biosecurity implications make deployments sensitive (see the Kinetics 2025 productivity report).

Healthcare prioritises admin automation and ambient documentation to ease clinician load, with cautious, safety‑first pilots in radiology and pathology because clinical missteps carry real harm.

Financial services and retail push real‑time fraud detection, automated lending and chatbots for efficiency, yet algorithmic pricing, discrimination and competition risks are flagged under existing laws and the new strategy's light‑touch approach.

Public sector priorities centre on fraud detection, service triage and infrastructure monitoring, balanced by the Government's Responsible AI Guidance and national AI Strategy which stress transparency, human‑in‑the‑loop controls and reliance on existing legal frameworks to manage privacy and consumer risk (read the New Zealand national AI Strategy overview).

In short: chase productivity where AI fits, but treat patient safety, privacy, competition and procurement choices as high‑risk priorities requiring strong governance and upskilling.

Infrastructure and investment are the twin engines behind New Zealand's AI opportunity: government strategy and industry analysis point to very large gains - estimates vary (industry-led analysis cited by the Global Government Forum suggests generative AI could add about NZ$76 billion to GDP by 2038, while a sector report from Wiise models a NZ$16 billion uplift by 2038), but all agree the prize is substantial and conditional on building capacity, not just policy.

Practical levers are already in motion - the Research & Development Tax Incentive and around NZ$611 million of AI‑related RDTI‑backed spend since 2019 provide fiscal support for innovation, and billions in data‑centre investment are positioning New Zealand as an APAC hub with lower latency and stronger data‑sovereignty options for public and private projects.

To turn potential into reality, the country needs modern cloud stacks, cleaner operational data and faster regional connectivity so rural agritech and urban health pilots scale; without that digital backbone the gains remain theoretical, not transformational.

For a snapshot of adoption and productivity trends see the Kinetics report, and for divergent economic estimates read the Wiise analysis and coverage in the Global Government Forum.

“Adopting generative AI alone could add NZ$76bn (US$45bn) to the New Zealand economy by 2038.”

The takeaway for beginners in New Zealand government is practical: treat AI as a tool to solve a defined problem, not a technology testbed - start with a small, governed pilot that aligns with the Public Service AI Framework and MBIE's Responsible AI guidance, map data flows to the Privacy Act, run an Algorithm/Privacy Impact Assessment, and require senior sign‑off and human‑in‑the‑loop controls before scaling; this low‑risk, iterative approach fits the government's light‑touch, principles‑based strategy and the national ambition that generative AI could add as much as NZ$76 billion by 2038 (see Nemko's summary of the Strategy).

Close the obvious skills gap by pairing pilots with focused training so teams can own procurement, bias checks and monitoring - practical courses such as Nucamp's AI Essentials for Work (15 weeks) teach prompt writing and workplace AI skills to turn policy into productivity.

Keep experiments small, document decisions, and use the official guidance as a checklist so New Zealanders reap efficiency gains while retaining public trust; MBIE's governance guidance is a useful place to start.

“The time has come for New Zealand to get moving on AI.” - Shane Reti